NVIDIA Delays Kyber Rack-Scale Architecture to 2028 Amid Manufacturing Challenges
NVIDIA has pushed back the launch of its Kyber rack-scale architecture—designed to house its 2027 Vera Rubin Ultra chips—by more than 12 months to 2028, according to research firm SemiAnalysis. The delay, attributed to manufacturing difficulties with a critical circuit board, has raised concerns about the AI giant’s ability to maintain its rapid product development pace, according to a report published on Monday.
What Is the Kyber Architecture, and Why Does It Matter?
The Kyber system is a server cabinet capable of housing 144 of NVIDIA’s most powerful chips in a single unit, enabling them to function as a unified high-performance computer. This design, which uses vertical compute trays to reduce latency, was originally slated to debut alongside the Vera Rubin Ultra chips in 2027. The architecture aimed to provide AI companies with the computational power needed to train and deploy advanced models, according to SemiAnalysis.
The delay stems from challenges in manufacturing the Kyber NVL144’s “PCB midplane,” a multi-layer printed circuit board that connects electronic modules within the system. “The PCB midplane remains challenging from a manufacturability standpoint,” SemiAnalysis stated, adding that a larger variant, NVL576, may also face delays or limited production. NVIDIA did not respond to requests for comment from CNBC.
How Has the Delay Impacted NVIDIA’s Product Roadmap?
The setback compounds existing pressures on NVIDIA’s product lines, as the company’s aggressive annual release schedule appears to be clashing with manufacturing constraints. A previously proposed workaround—combining two of NVIDIA’s current-generation racks to achieve similar performance—was scrapped after cloud service providers (CSPs) criticized the design as impractical and costly to operate, SemiAnalysis reported.
“It has since been cancelled due to heavy pushback from CSPs and hyperscalers over its odd design and heavy operational burden,” the research firm said. This leaves NVIDIA “without a proven solution to expand the scale-up world size for Rubin Ultra,” according to SemiAnalysis, potentially creating an opening for rivals like Advanced Micro Devices (AMD) and Google, whose in-house chips are already gaining traction with top AI labs.
What Does This Mean for NVIDIA’s Financial Outlook?
Despite the delays, SemiAnalysis projects that NVIDIA’s data-center compute revenue will exceed Wall Street expectations by 20% in the second half of fiscal 2027. The company’s current-generation Rubin systems are already in full production and set to ship to eight cloud partners, including Amazon Web Services, Microsoft Azure, and Google Cloud, starting this fall.
NVIDIA’s stock fluctuated in premarket trading on Tuesday, closing down less than 0.1% at $194.79, according to CNBC.
Why This Delay Matters for the AI Industry
The Kyber delay underscores the growing complexities of scaling AI infrastructure. As demand for high-performance computing surges, companies like NVIDIA face the dual challenge of innovation and manufacturability.
What’s Next for NVIDIA?
With the Kyber architecture now on track for a 2028 launch, NVIDIA will need to address manufacturing bottlenecks while maintaining its competitive edge. The company’s ability to deliver on its roadmap will be closely watched by investors and industry analysts alike. For now, the focus remains on the current Rubin systems, which are expected to drive revenue growth in the near term.